Backlog refinement:
turn feedback into work ready to plan.

Review customer requests, merge matching ideas and clarify the requirements. Give your team a prioritized backlog it can discuss, estimate and turn into a release plan.

Refine your product backlog

Follow a request through backlog grooming →

What is backlog refinement (grooming)?

// practical guide

An ongoing review of your product backlog

Backlog refinement is the ongoing work of reviewing, clarifying and ordering product backlog items. Backlog grooming is an older name for the same activity. It prepares selected work for a planning decision.

// the outcome

Know what is ready and what still needs an answer.

Keep the customer problem, scope, acceptance checks, estimate and dependencies clear. Separate reviewed priorities from requests that still need evidence or clarification.

// illustrative feedback-to-planning workflow

From scattered feedback to a refined backlog item

Acme receives Custom roles in Slack and a Team permissions request through its widget. The team confirms the same need, merges the matching Ideas and refines core roles for paid Acme workspaces.

  1. Review feedback: read the customer problem, source and useful context.
  2. Review duplicates: confirm the same need, choose the Idea to keep and merge the matching request.
  3. Clarify requirements: define the audience, scope, exclusions and acceptance checks; ask about missing details.
  4. Connect the backlog: link the reviewed Idea to Linear ENG-142 or create new work if needed.
  5. Prioritize: evaluate impact and effort, discuss scoring differences and review the ranking.
  6. Prepare the plan: choose work that is ready, then set dates, owners and dependencies for the release.
┌─ READY FOR RELEASE PLANNING ─────┐
│  selected          Custom roles  │
│  issue                  ENG-142  │
├──────────────────────────────────┤
│  Set dates and owners in Gantt   │
│  Connect task dependencies       │
│  Review the release window       │
│  Announce after actual release   │
└──────────────────────────────────┘
Illustrative requests, requirements and issue ID. The team reviews merges, scope and priority before planning. No production data is changed by this demo.
Illustrative requests, requirements and issue ID. Refinement prepares work for the team’s planning decision; a request is not automatically scheduled or released.
// review meaning · keep customer evidence

Review feedback and merge the requests that describe the same need

Bring requests from your widget, conversations and support into one place. Read the actual problem and customer context before deciding that two Ideas are duplicates.

In the sample, Custom roles and Team permissions both ask for configurable teammate access. The SSO request describes a different need and stays separate.

Open the public Idea card, choose Merge ideas and select the Idea that should remain. Voters, votes and tags move to that Idea. The original entries remain listed under Merged ideas, and you can unmerge them.

Keep distinct requirements separate even when the titles look similar. Merging public Ideas is separate from linking them to existing backlog work; it does not merge Jira or Linear issues.

Organize feature requests → · Collect feedback in your product → · Capture Slack requests →

┌─ REVIEW BEFORE MERGING IDEAS ────┐
│  Same need: custom team roles    │
│  keep              Custom roles  │
├──────────────────────────────────┤
│  Merge reviewed duplicate        │
│  Voters, votes and tags move     │
│  Keep SSO as a separate idea     │
│  Unmerge if the match is wrong   │
└──────────────────────────────────┘
Illustrative requests, requirements and issue ID. The team reviews merges, scope and priority before planning. No production data is changed by this demo.
// problem · scope · acceptance checks

Turn a vague feature request into requirements the team can review

Write down who needs the improvement, what they are trying to do and what evidence supports it. Define the scope and the questions you still need to answer.

For Custom roles, the sample scope is core roles for paid Acme workspaces. Acceptance checks include an admin creating a role with chosen permissions, assigning it to a teammate and verifying that the teammate can perform allowed actions while restricted actions remain unavailable. Guest access is out of scope.

Use an Idea template to prompt customers for clearer input. Ask the built-in Ducalis AI chat or your own assistant through cloud MCP to propose a requirements draft from the supplied feedback and source links; your team checks the facts, scope and acceptance checks before using it.

Use Questions when an issue or criterion needs clarification. Discussions are available on the Questions page and in the issue card. Keep unresolved questions visible before treating the work as ready.

Link the reviewed Idea to an existing issue or use + New to create work in Ducalis or your connected tracker. Put the reviewed scope and checks in the issue and retain the link to the customer Idea.

Draft requirements, stories and acceptance criteria with AI → · Configure a request template → · Choose your AI workflow →

┌─ LINK THE REVIEWED REQUEST ──────┐
│  idea              Custom roles  │
│  issue           Linear ENG-142  │
├──────────────────────────────────┤
│  Link existing or create new     │
│  Put scope and checks in issue   │
│  Keep the source idea linked     │
└──────────────────────────────────┘
Illustrative requests, requirements and issue ID. The team reviews merges, scope and priority before planning. No production data is changed by this demo.
// team evaluation · explicit trade-offs

Prioritize the refined work with impact, effort and customer context

Agree on the criterion definitions and rating scales. Assign evaluators with the relevant expertise and compare the refined request with bugs, improvements and other planned work.

Check effort, dependencies and uncertainty for the clarified scope. Use Criteria Alignment and Users Alignment to identify contrasting evaluations, then discuss the assumptions behind them.

Evaluators can revisit their estimates in the issue card. Review Final Scores under your configured permissions and decide what the team is ready to take into work.

A new estimate or a changed scope can change the ranking. Keep the decision rationale with the task and revisit it when new evidence arrives.

Prioritize customer requests and backlog work → · Review team scoring disagreements →

┌─ REVIEW BACKLOG PRIORITIES ──────┐
│  Recalculate configured model    │
│  Review ranking and trade-offs   │
├──────────────────────────────────┤
│  selected          Custom roles  │
│  issue                  ENG-142  │
│  scope      core roles + access  │
│  Team agrees what to plan next   │
└──────────────────────────────────┘
Sample people, scores and report percentages. Percentages illustrate the report, not a calculation from the sample ratings. The team makes the priority and release decisions.

Run backlog refinement asynchronously, then discuss the open questions

Prepare a focused set of items and give evaluators time to read the context. Bring the unresolved scope and scoring questions into the discussion.

Prepare the work and roles

Choose the items for this refinement cycle, define the criteria and assign evaluators. Product clarifies the outcome; customer-facing teammates supply evidence; engineering reviews effort and dependencies.

Assign users and criteria →

Collect evaluations and questions

Agree an evaluation window. Teammates score their assigned criteria and raise questions. If you use Evaluation Poker, configure private scores until Reveal, then review the results.

Configure Evaluation Poker →

Review the decision together

Discuss unclear requirements and contrasting estimates, revise scope or scores where needed and review selected work. Repeat refinement as new requests or evidence arrive.

Evaluate and re-evaluate issues →

Backlog refinement checklist: is this item ready to plan?

Use this practical checklist for the selected work. Your team defines the readiness standard for its product and release.

Review areaWhat the team can explain
Customer problemWho needs the change, why it matters and which source requests support it.
Duplicates and linksMatching Ideas have been reviewed; the surviving request is linked to the right work.
RequirementsScope, exclusions, acceptance checks and unresolved questions are visible.
Effort and dependenciesThe team has reviewed the estimate, uncertainty and work that must happen first.
Priority decisionThe configured scores and trade-offs have been reviewed; selected work has a clear rationale.
Planning inputSelected items are ready for the team to assign dates and owners and review the release window.
// refined backlog → scheduled work → available release

Bring the prioritized backlog into release planning

Choose the refined items your team will take into work. For Custom roles, plan access-model changes, the roles interface and release QA in Gantt.

Set dates and owners, connect dependencies and review what fits. If a predecessor takes longer, Finish to Start can shift dependent work and the team reviews the release window.

Choose which customer-facing Ideas to share on the public roadmap. Configure linked issue status rules separately, then review and approve the release announcement when the feature is available.

Plan dates and dependencies → · Show customers the progress → · Publish reviewed release updates →

┌─ REVIEW THE RELEASE PLAN ──────┐
│  scope           Custom roles  │
│  QA end                Oct 30  │
│  target       Nov 2 · planned  │
├────────────────────────────────┤
│  Team reviews the new window   │
│  Share updated scope + timing  │
│  Publish after actual release  │
└────────────────────────────────┘
Sample scores, people and dates. Finish to Start shifts dependent work; the team reviews the release target. Gantt dates and dependencies stay in Ducalis.

Prepare a backlog your team can evaluate

Use a cleanup playbook and checklists to review unclear items and keep the original problem visible.

Refine one customer request into work your team can plan.

Review the source, confirm duplicates, clarify the requirements and discuss the priority. Use that decision to prepare your next release plan.

Refine your product backlog

Use the refinement checklist → · See the full feedback-to-release workflow →